Make new replication slot test code less racy

alvherre · alvherre · commit 7099ba058035 · 2021-07-17T13:19:17.000-04:00
The new test code added in ead9e51 is racy -- it hinges on shared-memory state, which changes before the WARNING message is logged. Put it the other way around. Backpatch to 13. Author: Álvaro Herrera <alvherre@alvh.no-ip.org> Discussion: https://postgr.es/m/202107161809.zclasccpfcg3@alvherre.pgsql
diff --git a/src/test/recovery/t/019_replslot_limit.pl b/src/test/recovery/t/019_replslot_limit.pl
@@ -174,18 +174,21 @@
 my $logstart = get_log_size($node_master);
 advance_wal($node_master, 7);
 
-# This slot should be broken, wait for that to happen
-$node_master->poll_query_until(
-	'postgres',
-	qq[SELECT wal_status = 'lost' FROM pg_replication_slots
-	WHERE slot_name = 'rep1']);
-
-# WARNING should be issued
-ok( find_in_log(
-		$node_master,
-		"invalidating slot \"rep1\" because its restart_lsn [0-9A-F/]+ exceeds max_slot_wal_keep_size",
-		$logstart),
-	'check that the warning is logged');
+# wait until the WARNING is issued
+my $invalidated = 0;
+for (my $i = 0; $i < 10000; $i++)
+{
+	if (find_in_log(
+			$node_master,
+			"invalidating slot \"rep1\" because its restart_lsn [0-9A-F/]+ exceeds max_slot_wal_keep_size",
+			$logstart))
+	{
+		$invalidated = 1;
+		last;
+	}
+	usleep(100_000);
+}
+ok($invalidated, 'check that slot invalidation has been logged');
 
 $result = $node_master->safe_psql(
 	'postgres',
